The need to remain cost and performance competitive in the microelectronics industry has caused continually increasing semiconductor device density in integrated circuits. To facilitate the increase in device density, new technologies are constantly needed to allow the minimum feature size of these semiconductor devices to be reduced. This requires high-resolution lithography, the principal technique used in patterning microelectronics circuitry. Over approximately the last 20 years, the industry has migrated to shorter wavelength photolithography as the primary means of scaling the resolution to sustain the progressive demand for smaller features. The wavelength of photolithography has migrated from mid-ultraviolet (MUV) wavelengths (350-450 nm) to deep-UV (DUV) radiation (190-300 nm) and vacuum UV (VUV, 125-160 nm). Likewise the photosensitive materials used in photolithography, resists, have evolved. MUV lithography employed diazonaphthoquinone (DNQ) and novolac-based resists. These materials offered high performance but were not extendible to DUV and VUV wavelengths due to their opacity at these shorter wavelengths. In addition, these resists were not of sufficient sensitivity to afford high throughput manufacturing.
In response to the need for new, lower opacity, higher sensitivity materials for DUV imaging, Ito et al. disclosed (U.S. Pat. No. 4,491,628) the development of chemically amplified resists (CARs) based on the photochemically-generated-acid (from a photosensitive acid generator) catalyzed deprotection of an acid-labile polymer. That is, for positive tone CARs, acid labile moieties of the polymer are cleaved by an acid-catalyzed thermolysis reaction that renders the deprotected form of the polymer soluble in a subsequently applied developer, such as an aqueous base. Thus, an image of the projected patternwise radiation is formed in the resist film after development, which can then serve as an etch-resistant mask for subsequent pattern transfer steps. The resolution obtained is dependent on quality of aerial image and ability of resist to maintain that image.
CARs have been developed for 248, 193, and 157 nm lithography. The resolution, R, of an optical projection system such as a lithography stepper is limited by parameters described in Raleigh's equation:R=kλ/NA,where λ represents the wavelength of the light source used in the projection system and NA represents the numerical aperture of the projection optics used. “k” represents a factor describing how well a combined lithography system can utilize the theoretical resolution limit in practice and can range from about 0.8 down to about 0.5 for standard exposure systems. The theoretical dimensional limit of equal-sized half-pitch features is one quarter of the wavelength, λ. (k =0.25) when NA =1, as the dose applied to the resist is equal to the square of the intensity, and thus the resolution cannot be modulated by any more than λ/4, or a pitch of λ/2. The resolution attainable with each advancing generation of materials has been extended toward these limits through the use of low k techniques and high numerical aperture tools. For 157 nm lithography, the latest VUV wavelength being developed for manufacturing, and using a very high but potentially manufacturable NA of 0.95, the minimum k factor (i.e. λ/4) equals ˜40 nm. To obtain images below this feature size, either an extension of NA to >1, such as is afforded with immersion lithography, or with a non-diffraction limited, non-optical lithography system, such as the so-called next generation lithography (NGL), are options. The most promising of these NGLs are extreme ultraviolet (EUV, sometimes referred to as soft x-ray) or electron beam lithography (EBL).
As the desired feature size decreases, the resolution capability of many current resists is not sufficient to yield the smaller features. The need to achieve less than 100 nm resolution has prompted a push toward increasing numerical aperture (NA) exposure tools. The higher NA allows for improved resolution of smaller feature sizes, however, the higher NA also reduces the depth of focus of aerial images projected onto the resist film. Since the depth of focus is more shallow, the thickness of the resist film becomes a factor in properly exposing the resist. Thus, thinner resist films may be required for proper exposure at high resolution, but such films often do not yield acceptable overall performance, especially when considering etch requirements for the underlying substrate.
As the resist film is thinned to account for the higher NA, the resist becomes less suitable as an etch mask against later processing of the underlying semiconductor substrate. For example, since the resist film is thin, variation in thickness becomes more significant and may introduce defects into subsequent devices formed on the substrate. Also, micro-channels often form in the upper portions of a resist layer during transfer of the resist image to the substrate by etching. When the resist layer is thin, the micro-channels may extend to the underlying substrate rendering the resist less effective as a mask.
In addition, the process latitude of many current resists is not sufficient to consistently produce the smaller desired features within specified tolerances. Some of the process parameters where variance may be difficult to avoid include bake time and temperature, exposure time and source output, aerial image focus, and develop time and temperature. The process latitude of a resist is an indication of how wide such variations can be without resulting in a change in the resolution and/or image profile (i.e., size and/or shape of a resist image). That is, if the process latitude is sufficiently wide, then a process parameter may vary, but the variance will not produce a change in the resist image incompatible with specified tolerances.
One approach that enables the use of higher NA exposure tools as well as a thinner resist film is multilayer resist processing. One type of multilayer resist processing uses a bilayer (two layer) imaging scheme by first casting a highly energy absorbing underlayer on the semiconductor substrate then casting a thin, silicon-containing imaging layer (resist film) on top of the underlayer. The silicon-containing resist provides good etch selectivity for anisotropic dry etch processing, such as reactive ion etch (RIE) using an oxygen-containing plasma. Next, selected portions of the silicon-containing layer are exposed and developed to remove the unexposed portions of a negative resist film or the exposed portions of a positive resist film. Generally, the underlayer is highly absorbing at the imaging wavelength and is compatible with the imaging layer. Interactions to be considered include adhesion between the two layers, intermixing, and contamination of the imaging layer by the components of the underlayer. Also, the refractive index of the underlayer is matched to the refractive index of the silicon-containing resist layer to avoid degrading the resolution capability of the silicon-containing resist.
Conventional underlayers include diazonapthoquinone (DNQ)/novolac resist material or novolac resin cast on the semiconductor substrate. The underlayers are typically selected to have good selectivity for an anisotropic etch, and to be sufficiently rigid to minimize feature collapse.
For the imaging layer, resists containing a wide variety of silicon-containing polymers have been used, including silsesquioxane, silicon-containing acrylics, silanes, etc. Among the several possible silicon-containing polymers, aqueous base-soluble silsesquioxane polymers, such as poly(p-hydroxybenzylsilsesquioxane) (PHBS), have emerged as the most promising candidates for silicon-containing polymers in bilayer resist systems. Unfortunately, although it is promising, PHBS suffers from several shortcomings.
A phenomenon known as image blur diminishes the integrity of the pattern (Hinsberg et al., Proc. SPIE, 2000, 3999, 148 and Houle et al., J. Vac. Sci. Technol B, 2000, 18, 1874). Image blur can be defined as the deviation of the developable image from that of projected aerial image which is transferred into the film as the concentration of photochemically generated acid. While accelerating the rate of the deprotection reaction, the application of thermal energy diminishes the fidelity of the aerial image of acid formed during the patternwise exposure. Image blur can be divided into two contributing factors: gradient-driven acid diffusion and reaction propagation. Both factors contribute to blur, but to different degrees and with different temperature dependence.
The first factor contributing to image blur is often referred to as acid diffusion and can be described by Fickian diffusion models for solids (Hinsberg, 2000). Choice of the photoacid being generated from the photoacid generator (PAG) and the mobility in the chosen polymer matrix dictate this factor. The mobility in the polymer matrix is dependent on the comprising chemical functionality of the polymer, the free volume of the matrix, the glass transition temperature (Tg) of the polymer and the temperature and time of baking steps encountered during the resist processing. Processing at temperatures above Tg will tend to increase image blur.
A second contributing factor to image blur is sometimes described as reaction propagation (Hinsberg, 2000; Houle, 2000) and is best described by Arhenius behavior. Activation energy (enthalpy, hereinafter referred to as Ea), volatility of products (entropy), and the availability and concentration of deprotection-reaction-dependent co-reactants such as moisture dictate the degree to which the reaction propagates away from the original acid profile. Higher Ea for the deprotection reaction requires higher baking temperatures, which will tend to increase image blur. Current resists for 193 nm lithography have relatively high Ea (i.e. greater than about 20 kcal/mol).
In order to achieve high resolution, high sensitivity and high degree of process latitude, both image blur factors must be eliminated or minimized. Both of these contributing factors can be tempered by the addition of acid-quenchers, or bases, which have been shown to reduce image blur. Additionally it has been recognized that image blur is temperature dependent, and tends to increase as processing temperature increases. Breyta et al. disclose that appropriate baking conditions can optimize the resolution attainable with CARs in U.S. Pat. No. 6,227,546. However, since the extent of thermally induced image blur has been estimated to be on the order of 10-50 nm with conventional resist processing schemes by various researchers (Hinsberg 2000; Houle 2000; Krasnaperova et al., J. Vac. Sci. Technol. B, 1994, 12, 3900; Lin et al, Science 2002, 297, 372), processing that reduces this phenomenon further are desirable.
A higher Tg is often desired to avoid blur-inducing film instability. Higher Tg is typically achieved by providing longer cyclic chains in the side groups. For example, to increase Tg, bulky groups, such as alicyclic structures, are used in the side chain. Unfortunately, these bulky groups tend to decrease the transparency of the resist formulation at the wavelengths of interest. The addition of fluorine in the polymer structures can improve transparency, but in some instances the addition of fluorine may not be enough to achieve sufficient transparency and the other desired resist properties and performance.
Thus, there is a need for improved silsesquioxane polymers structures useful in resist compositions and in photolithography processes, that have improved transparency at wavelengths at or below 193 nm with minimized image blurring.